American Scientist

Alarms first sounded on September 18, 1988, when headlines in the Eugene Register-Guard, Oregon's second-largest newspaper, blared: "Decaying dam holds tide of trouble: It threatens to wash away town of 1,500." The article was about a structure the Army Corps of Engineers had built across Willow Creek (a tributary of the Columbia River) six years earlier, near the rural community of Heppner, using what was then a novel construction technique. The story ignited a political firestorm.

A few townspeople were frightfully aware of what such a disaster was like. As young children, they had survived the catastrophic Heppner flood of 1903, which took 251 lives. As fear turned to anger, some residents saw themselves as guinea pigs. Editorials in the local newspaper bristled with accusations. One read: "They were looking for a sparsely settled area to experiment with this new type of dam, and now we're stuck with it."

A New Type of Dam

The Flood Control Act of 1965 first authorized the building of the Willow Creek dam, which was to be located at a narrowing of the river, about 100 meters upstream of the Heppner city limit. To reduce the cost of the project (first estimated at $50 million) and to speed up construction, the Corps decided to use roller-compacted concrete, which had been touted in engineering journals during the 1970s as a revolutionary material suitable for, among other things, dam construction. Willow Creek thus became the first dam in the United States to be built in this way.

Ground was broken in May 1982. About 330,000 cubic meters of concrete were used, most of the material being laid down in successive layers, each 30 centimeters thick, which were then compacted with vibratory rollers. When completed six months later, the dam consisted of 160 layers, stacked to a height of nearly 50 meters, with a total length of some 550 meters. The dam's interior was accessible through a longitudinal maintenance tunnel 275 meters in length. By February 1983, construction was finished. Total cost proved to be around $35 million—well below the original price tag, especially if one takes inflation into account.

Corps engineers then proceeded to fill the reservoir and quickly discovered that water seeped through the dam at a significant rate, estimated at 11 cubic meters per minute, escaping mostly through the junctions between the concrete layers and coalescing into streamlets that cascaded down the dam face. To stem the flow, the engineers largely emptied the reservoir and injected cement grout into a series of vertical drill holes, each extending from the crest of the massive structure to its foundation. This effort, costing an additional $2 million, slowed the leakage to about 3 cubic meters per minute.

In May of 1984, while working as a Corps limnologist, I began a long-term study of the reservoir. What I found was a water body highly enriched with nitrogen and phosphorus and with dissolved and particulate organic matter, originating mostly from cattle feedlots and other agricultural sources in the watershed. As plant nutrients, these substances spurred the growth of enormous quantities of cyanobacteria (blue-green algae) and planktonic diatoms. After dying, these organisms settled to the bottom, where they decomposed along with other organic material washed in during thunderstorms and with the spring runoff from the surrounding snow-covered hills.

During summer, the reservoir became divided into two distinct zones, with warmer, less-dense water above a few meters depth and colder, denser water below. The density contrast prevented any significant circulation between the two layers. Consequently, microbial decomposition of organic matter along the bottom gradually depleted oxygen and greatly increased the carbon dioxide concentration deeper down. By mid-summer, the water that was more than a few meters below the surface became completely anoxic.

In the late summer of 1985, workers inspecting the maintenance tunnel reported a strong "rotten egg" smell, indicating the presence of hydrogen sulfide gas (which can be lethal in sufficient concentrations). Subsequent air testing detected not only H₂S, but also potentially explosive methane. Corps officials issued a warning, requiring people entering the tunnel to wear gas masks or other breathing devices.

In a January 1986 memorandum, I warned my Corps colleagues about the possibility that sulfur-oxidizing bacteria in reservoir waters might produce sufficient sulfuric acid to erode the concrete and weaken the dam's structural integrity. This process could also explain the persistent leakage. (I cited several studies of concrete corrosion by H₂S-derived sulfuric acid in municipal sewer systems.) I recommended that the Corps commission a thorough study of the dam by independent scientists and that it install an aeration system to replenish oxygen in the deeper waters behind the dam during summer months. On paper at least, one commercial aeration unit, costing $90,000, would be sufficient to preclude the formation of H₂S and other reduced chemicals.

My superiors at the Corps approved the independent scientific review but deferred on my proposal for aeration until further study could confirm that it would help in slowing deterioration of the concrete. In the summer of 1986, I assembled a team of outside scientists to assess the situation: four highly qualified researchers from Oregon State University and the University of Washington. Their investigation, conducted during the following two years, determined that chemosynthetic bacteria living in the interstices of the roller-compacted concrete and on the surface of the dam acidified the water seeping through. Chief among these problematic organisms was Thiobacillus, a bacterium capable of oxidizing H₂S to sulfuric acid. Other oxidizing bacteria converted ammonia to nitric acid. These acids, along with carbonic acid and organic acids already present in these waters, dissolved the calcium carbonate of the concrete and redistributed some of it within the dam structure. Another fraction was precipitated on the dam face, with the balance being released downstream. The scientific team estimated that, on average, 29 metric tons of cement materials were being carried away annually. This loss, they warned, could compromise the dam's structural integrity, making it unsafe.

Damage Control

In October 1988, shortly after the Eugene Register-Guard got wind of these results, Oregon's Senator Mark Hatfield became involved. At his behest, the Corps dispatched a special team of four senior-level engineers and a consulting concrete expert to Oregon to investigate. The engineers—hand-picked from headquarters staff in Washington, D.C.—worked for the Chief of the Directorate of Engineering and Construction, who had received a $10,000 Presidential award in 1987 for supervising the design and construction of Willow Creek dam, creating an obvious conflict of interest (which journalists and others involved somehow overlooked).

On arrival, the team spent some hours inspecting the dam before reviewing the scientific data. After just a few days of deliberation, the team boldly announced to the news media that "There is no reason for concern regarding the structural safety of the dam. The dam is indeed a reliably safe structure." The engineers then impugned the scientists' research, stating: "Criticism of the [roller-compacted concrete] design and construction concept is not warranted and without merit." And they killed my proposal for aeration by saying, "Despite the presence of hydrogen sulfide gas, there is no evidence that acid is wearing away the dam."

Senator Hatfield's office immediately issued a news release, which in part read: "Allegationsof health and safety problems at Willow Creek were quite serious. After a thorough study by Army experts, it appears reasonable to assert that these concerns were unfounded." On November 17, 1988, the Oregonian, Oregon's largest newspaper, published these declarations in a front-page story headlined: "Dam safe, experts agree."

Further efforts to end the debate quickly followed. Contrary opinion within the Corps was suppressed, and contracts with the outside scientists were not renewed, thus ending what to that point had been collaborative and productive research into the safety of the dam.

In July 1989, the scientists issued their final report, which detailed the deterioration of the roller-compacted concrete, but the Corps refused to release the document to the public. Journalists at the Eugene Register-Guard attempted to obtain a copy by filing a request under the terms of the Freedom of Information Act, but the Department of the Army's Assistant Chief Counsel for Litigation denied the request, claiming that release of the report "could discourage honest and frank discussions within the Corps." The Register-Guard decided not to appeal. The controversial document was, however, released in 1990, but it was appended to the Corps' own final report, which was clearly aimed to discredit the scientists' research methods and findings.

Lessons Learned

Were the scientists' concerns justified? It's difficult to say. Leakage has diminished considerably and is now estimated at less than 1 cubic meter per minute. Only patches of wetness are currently visible on the dam face, perhaps validating the engineers' claim that dams made of roller-compacted concrete eventually stop leaking because of their self-sealing tendency.

Still, confidence in the dam's infallibility within the Corps was never complete: Even after the official safety assessment was made, Corps engineers continued to monitor leakage flow and chemistry for the next 15 years at a cost of several million dollars. In 2004, an aeration system was finally installed. The Corps insists that its use is unrelated to dam safety and represents only a measure to control algal blooms by reducing the nutrient loading that results from deep-water anoxic conditions.

Thankfully, the dam didn't collapse, as I and others once feared it might, but the margin of safety under which it operates is not really well known. Indeed, in the two decades since the Willow Creek story unfolded, much has remained unclear. Yet some things are for sure: The public had a right to know about any anomalies in the functioning of the dam; the press was obliged to investigate; and Congress had a duty to protect local citizenry from a structure that it had put in place. But as the history of events shows, the process of technical study and public discourse went awry, beginning with the sensationalized news story, which naturally enough triggered a quick-fix Congressional demand for immediate answers. Under tremendous political pressure, the Corps scrapped a promising scientific investigation and, with only an abiding faith in roller-concrete construction to support its position, arbitrarily declared the dam "reliably safe," a strange phrase that perhaps reflects the fact that the verdict was not without uncertainty.

At that point, though, a more straightforward statement by the Corps that the dam could conceivably become unsafe would have likely required some dire—and discreditable—measures, the most drastic being a forced evacuation of the townspeople and even removal of the dam. So when the troubling news broke, the strategy the Corps adopted was not to admit to any possibility of risk. In a more rational world, all parties would have acknowledged that absolute safety would be impossible to guarantee instantly. The Corps could then have temporarily drained the reservoir to ease concerns and to allow a better examination, and it could have been open about its ongoing studies and efforts to ensure that any dangerous deterioration of the dam would be detected and remedied well before it had a significant chance of endangering lives or property.

I like to think that scientists and engineers faced with a similar problem today would tackle it in such a mature and sensible fashion. But I fear they might well repeat the mistakes made at Willow Creek a couple of decades ago, which is why it's useful to look back on the episode—not just as an amusing piece of history, but also for the (excuse the expression) concrete lessons it can provide to those people whose job it is to assess, communicate and manage risk.